Author: Russell Reagan
Date: 16:30:09 04/12/04
Go up one level in this thread
On April 12, 2004 at 18:22:18, Omid David Tabibi wrote: >On April 11, 2004 at 17:05:00, Dan Andersson wrote: > >> Any reasonable one should. But since C/C++ have serious holes in their type >>systems amongst other shorcomings it is usually in a limited form. There have >>been proposals of a special tail_call form. >> Earlier GCC (old 2.x branch) only performed it at the the front end. And only >>with an explicit return. Mutual recursion was also a hindrance. >> As an aside I believe that the MS C# doesn't do tail call optimizations. > >I was reading some benchmarks in a newsgroup suggesting that C# is even slower >than Java. So, as far as optimizations go, don't count on it :) I wouldn't believe anything about C# efficiency unless I tested the specific problem myself. It seems like every benchmark that I see comparing C/C++, C#, and Java tells a different story. Sometimes C# is reported to be only 2% slower than natively compiled C/C++. Sometimes it is reported to be horribly slower. Sometimes it is reported to be significantly faster than Java, and sometimes the opposite. With C/C++ we can usually make a decent educated guess as to how something will perform, and if you want to be sure: try it and find out. With C#, I think the answer is always: try it and find out. I have started learning C#, and so far I really like it. It makes things that are sometimes awkward in C/C++ seem very clean and easy. Plus, it has a huge library. It would be very easy to write a Winboard/UCI interface in C#, since it has things like text pipe interprocess communication, multithreading, GUI development, and network communication built into the language. If C# was only 2% slower than natively compiled C/C++, I'd kiss C/C++ goodbye (I'm not holding my breath though).
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.